Skip to main content

stackprep-pro

stackprep-pro — interview & certification prep MCP server for any AI client

Works with any MCP-compatible client — Claude Code, Cursor, Cline, Windsurf, Continue.dev, Codex CLI, and any other client that supports the Model Context Protocol. No API key required — your existing AI subscription does the work.

Available on PyPI: uvx stackprep-pro


What it does

stackprep-pro is a pure state-management MCP server. It tracks your session and study packs on disk; your AI client (Claude, Cursor, Codex, etc.) handles all the question generation and scoring logic using the skill rules returned at session start.

  • One question at a time — interview or certification mode
  • Continuous flow — the next question follows every answer automatically, no "continue? (y/n)" prompts
  • Instant scoring with doc links after every answer
  • Auto-detects wrong/partial answers and builds a named study pack
  • Sessions and study packs saved to disk — resume anytime, sync via iCloud
  • Resume in-progress sessions across conversations or devices

Install

uvx stackprep-pro

Requires uv. Install it with curl -LsSf https://astral.sh/uv/install.sh | sh.


Configure your MCP client

The config is the same for every client — just point to uvx stackprep-pro. No API keys, no authentication, no accounts — stackprep stores everything as plain files on your own machine.

Prerequisite: install uv (it provides uvx):

curl -LsSf https://astral.sh/uv/install.sh | sh

Claude Code

Recommended — register it globally so it works from any directory (the normal way you'd use it):

claude mcp add stackprep --scope user -- uvx stackprep-pro

Then just run claude from anywhere. Claude Code works as normal — stackprep stays out of the way until you want it. To start a prep session, type stackprep-pro (or ask to prep for an interview or certification). You can also register it per-project instead of globally; both work and can coexist.

Alternative: per-project config

If you'd rather scope it to a single project, create .mcp.json in that project's root instead:

{
  "mcpServers": {
    "stackprep": {
      "command": "uvx",
      "args": ["stackprep-pro"]
    }
  }
}

Cursor

Create ~/.cursor/mcp.json (global — works from any directory):

{
  "mcpServers": {
    "stackprep": {
      "command": "uvx",
      "args": ["stackprep-pro"]
    }
  }
}

Then open Cursor → Cmd+Shift+J → MCP tab — stackprep should appear with a green dot.

Codex CLI

Add to ~/.codex/config.yaml:

mcpServers:
  stackprep:
    command: uvx
    args:
      - stackprep-pro

Any other MCP-compatible client

The pattern is always the same:

{
  "mcpServers": {
    "stackprep": {
      "command": "uvx",
      "args": ["stackprep-pro"]
    }
  }
}

Paste this into whatever config format your client uses (Cline, Windsurf, Continue.dev, etc.).


How a session works

Once a session starts, it keeps going until you end it — there is nothing to confirm between questions.

  • Answer — just type your answer (a letter or free text). You get the result, a short explanation and a docs link, and the next question follows in the same message.
  • end — type end at any time to finish. You'll be asked whether to save the session to continue later (you name it), and whether to save a study pack (you name it).
  • S — save a study pack mid-session without ending it.
  • Auto-save — progress is saved after every answer, so closing the window abruptly never loses your session.

Use end rather than X to finish — X closes some CLIs outright.

Every question shows a reminder footer: Reply with your answer to continue · type end to finish · S to save a study pack.


Study pack storage

Study packs and sessions are saved to ~/.stackprep/ by default.

Sync across devices with iCloud (recommended on macOS):

# Add to ~/.zshrc or ~/.zprofile
export STACKPREP_PACKS_DIR="$HOME/Documents/stackprep-packs"

~/Documents is synced to iCloud by default on macOS (requires iCloud Drive > Desktop & Documents enabled). Your packs will be available on any Mac signed into your Apple ID — and readable via the Files app on iPhone.

Custom path:

export STACKPREP_PACKS_DIR="/path/to/your/folder"

Point this at any Dropbox, Google Drive, or OneDrive folder for cross-platform sync.


Environment variables

Variable Default Description
STACKPREP_PACKS_DIR ~/.stackprep Root directory for packs and sessions.

Skills (modes)

Mode Description
certification description: Certification prep skill for the stackprep-pro MCP server. Activated when mode is "certification". Drives question generation, scoring, adaptive difficulty, and study pack creation.
interview description: Interview prep skill for the stackprep-pro MCP server. Activated when mode is "interview". Drives question generation, scoring, adaptive difficulty, and study pack creation.

Tools

Tool Description Args
begin Start a stackprep prep session. Call this ONLY when the user explicitly triggers stackprep — i.e.
start_session Start a new stackprep session. Returns a session ID and the skill rules for the AI to follow. mode, cert_name, cv, jd, extra_topics
submit_answer Record the result of an answered question. session_id, result, question
flag_for_study Manually flag the current question for the study pack. session_id, question
save_session Save an in-progress session so the user can continue it later. session_id, session_name
exit_session ALWAYS call this the moment the user wants to leave a session in progress — any of: "exit", "quit", session_id
discard_session Permanently delete a session. Call this when the user is exiting and answers NO to saving the session_id
end_session End the session. Returns the score and flagged topics so the AI can generate a study plan and study pack. session_id
save_study_pack Save the study pack content to disk. session_id, name, content
list_sessions List saved sessions. Call this silently when the user wants to continue. Never mention this tool to the user. mode
resume_session Resume a previously saved session. Returns full session state and skill rules. session_id
list_study_packs List saved study packs. Call this silently when the user wants to see or load a study pack. Never mention this tool to the user. mode
load_study_pack Load a previously saved study pack by name. name

Session flow

Certification:

list_sessions()                                          ← always called first
→ start_session(mode="certification", cert_name="AWS SAA-C03")
→ [AI generates questions one at a time]
→ submit_answer(session_id, result="correct"|"partial"|"incorrect", question="...")
→ ... repeat ...
→ end_session(session_id)
→ save_study_pack(session_id, name="aws-saa-week1", content="...")

Interview:

list_sessions()                                          ← always called first
→ start_session(mode="interview", cv="...", jd="...")
→ [AI generates questions one at a time]
→ submit_answer(session_id, result="correct"|"partial"|"incorrect", question="...")
→ ... repeat ...
→ end_session(session_id)
→ save_study_pack(session_id, name="python-interview-june", content="...")

Resuming a session:

list_sessions()                → shows in-progress sessions
→ resume_session(session_id)  → loads state + skill rules, continues where you left off

Loading a saved study pack:

list_study_packs()
→ load_study_pack(name="aws-saa-week1")

Session persistence

Every session is saved to disk on every update. At the start of each new conversation the AI automatically calls list_sessions and asks whether you want to resume an in-progress session or start a new one. Sessions are stored in ~/.stackprep/sessions/ (or your custom STACKPREP_PACKS_DIR).


Also available as a Claude Code plugin

For Claude Projects or direct Claude.ai use, the behaviour rules are also available as a standalone skill file at plugins/stackprep — no install needed.


Contributing / Development

git clone https://github.com/youngpada1/stackprep-pro
cd stackprep-pro

# Install dependencies
uv sync

# Activate the pre-commit hook (auto-regenerates README on every commit)
git config core.hooksPath .githooks

# Run the server locally
uv run stackprep-pro

The README is auto-generated from server.py tool definitions and the skills files in src/stackprep_pro/skills/. To regenerate manually:

uv run python scripts/generate_readme.py

License

MIT — Flavia Fauconnet

Also Available in MCP Rush ↴

Stackprep Pro MCP server

Metadata

Release files for stackprep-pro 0.2.30

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for stackprep-pro 0.2.30
File Size Uploaded Provenance Actions
stackprep_pro-0.2.30.tar.gz 64.3 kB Without provenance Details for stackprep_pro-0.2.30.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for stackprep-pro 0.2.30
File Interpreter ABI Platform Actions
stackprep_pro-0.2.30-py3-none-any.whl Python 3 none any Details for stackprep_pro-0.2.30-py3-none-any.whl

Total release size: 83.2 kB

Release files / stackprep_pro-0.2.30.tar.gz

Download URL stackprep_pro-0.2.30.tar.gz
Size 64.3 kB
Tags Source
SHA-256 checksum
How to use checksums
44a11e0a61b140dd1b254d86a110a8def83b9770f29fb6d3f70a489fd72e8d3d
BLAKE2b-256 checksum
How to use checksums
3488b28e25670192adf9eeecfaf2d3f422b4c364561134a88cda5f2824b3d1cc
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Release files / stackprep_pro-0.2.30-py3-none-any.whl

Download URL stackprep_pro-0.2.30-py3-none-any.whl
Size 18.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
78ea897bd794b62f10dcb7e9e31045ad4a411c3a2a5c3d37699c3196644b9094
BLAKE2b-256 checksum
How to use checksums
39aecd2efddacaa7bc8e665ae07d031a49478ddb9d18b273465c10cda72c5953
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Security for stackprep-pro 0.2.30

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.2.30 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page